Hearing, Balance and Communication, 2013; Early Online: 1–9
ORIGINAL ARTICLE
Hearing, Balance and Communication Downloaded from informahealthcare.com by 147.122.60.69 on 07/23/13
For personal use only.
Language development in infants: What do humans hear in
the first months of life?
ALAN LANGUS & MARINA NESPOR
International School for Advanced Studies (SISSA), Language, Cognition and Development Laboratory – Trieste, Italy
Abstract
In this article we discuss experimental work on language acquisition that we have carried out in recent years. We first
describe the most common methods used in infant research, then we concentrate on linguistic rhythm and the aspects of
language that might be learned in the first year of life on the basis of signals contained in the speech stream. These include
the basic level of rhythm carried by the two most basic phonological categories of consonants and vowels, and rhythmic
alternation at the phrase level and its signal to syntax. Because linguistic rhythm is one of the first aspects of language that
infants perceive and represent, the research discussed may help to diagnose hearing problems and lead to new ways of
training individuals with speech impairments.
Key words: language acquisition, infants’ research, rhythm, rhythmic classes, Iambic-Trochaic Law
Introduction
Rhythm pervades the universe: natural as well as
cultural phenomena are governed by rhythm. The
waves of the sea move rhythmically. Day and night
alternate rhythmically. Our hearts beat rhythmically,
as do those of other animals. We breathe rhythmically
and we walk rhythmically. Music in all cultures is
governed by rhythm and so is dance, but what is
rhythm? To our knowledge, the most essential – thus
more general – definition of rhythm can be found in
“Rhythm is order in movement, Plato, The Laws,
Book II, 93”. While the elements that establish order
may vary in different natural and cultural phenomena, order is always established by alternation, or the
regular recurrence of a pattern.
What is rhythm in language? What are the elements that establish order in the movement of speech?
Linguistic rhythm, as rhythm in music, is hierarchical
in nature. There are different elements that establish
rhythmic alternation – or order – at different levels
of the rhythmic hierarchy (1,2). For example, on the
most basic level, linguistic rhythm is signalled by
the space occupied by vowels in the speech stream
(V%) and the standard deviation of consonantal
intervals (∆C) (3). However, if we move higher on
the prosodic hierarchy – to the phonological phrase
level – we find that rhythm is signalled through the
systematic recurrence of prosodic cues such as pitch,
intensity or duration in the prominent elements of
phonological phrases.
Because of infants’ sensitivity to prosody, and
because the alternation of rhythm at different levels
of the rhythmic hierarchy signals different linguistic
properties, it offers an ideal opportunity for investigating the early stages of language acquisition.
The study of rhythm perception in early infancy
can therefore not only tell us whether infants can
discriminate languages on the basis of rhythm,
but also when infants become sensitive to the
specific cues that are carried by rhythm. This is
particularly relevant, since rhythmic alternation has
been shown to offer cues to the size of the syllabic
repertoire (3), to the mean length of words (4) and
to basic syntactic properties of language (5,6).
Since the sound pattern of a language, in particular
its prosody, gives cues to different grammatical
properties, it is feasible that infants start their
first steps into the acquisition of grammar at the
pre-lexical stage, i.e. before knowing the meaning
of words.
Correspondence: M. Nespor, ERC – PASCAL, SISSA via Bonomea 265, 34136 Trieste – Italy. E-mail: [email protected]
(Accepted 17 June 2013)
ISSN 2169-5717 print/ISSN 2169-5725 online © 2013 Informa Healthcare
DOI: 10.3109/21695717.2013.817133
2
A. Langus & M. Nespor
Hearing, Balance and Communication Downloaded from informahealthcare.com by 147.122.60.69 on 07/23/13
For personal use only.
To give the reader a flavour of infant research on
language acquisition, we therefore begin with a brief
description of the various techniques used to test
newborns and young infants. We then concentrate on
what infants know about language when they are
born and how they progressively tune in to the language of their environment. This will serve as a basis
for discussing the aspects of linguistic rhythm infants
know, how infants process linguistic rhythm at different levels of the rhythmic hierarchy, and what they
learn about their language of exposure through
rhythm during the first year of life.
How do we know what infants know?
Unlike in experiments with human adults, preverbal
infants cannot be asked what they know or be told
to cooperate in specific tasks. Neither can they
be extensively trained like laboratory animals to
perform tasks that they intuitively do not understand.
It is important to bear in mind that infants’ voluntary
behaviour is highly limited: infants begin to intentionally grasp things around 5–6 months of age, utter
their first words around their first birthday and begin
to utter multi-word utterances around two years of
life. As a consequence, to investigate what infants
know before they can overtly tell us, researchers have
to rely on ingenious ways of interpreting and using
infants’ naturally occurring responses to environmental stimuli. In the following, we briefly describe
the arsenal of research methods ranging from behavioural methods that rely on infants’ behavioural
responses, such as sucking and looking, to neural
responses in the brain measured with imaging
techniques such as EEG and fNIRS.
Behavioural methods
Sucking is one of the first behavioural instincts of
infants. While nutritive sucking is essential for
infants to acquire food, non-nutritive sucking is
considered a natural reflex for infants to demonstrate their need for contact, to feel secure and
relaxed. As infants tend to suck more when presented with novel environmental stimuli than with
familiar ones (7), the reflex has extensively been
used to determine infants’ sensitivity to speech (8).
In a typical non-nutritive sucking experiment a
pacifier is held in the infant’s mouth. The pacifier
is connected to two loudspeakers that react to the
subject’s sucking so that sucking with enough energy
causes the loudspeakers to emit a linguistic sound,
for example ‘ta’. In the first phase of the experiment, called habituation, each high amplitude sucking triggers the same sound. After a while the
subjects suck less and when the sucking rate reaches
a habituation criterion (usually 50%), half of the
neonates – the experimental group – hears a sound
that differs minimally from the habituation sound,
for example ‘da’. The other half of the subjects – the
control group – keeps listening to the habituation
sound. If the new stimulus triggers high amplitude
sucking while the old one makes the subject suck
progressively less, it can be concluded that the
infant can distinguish already at birth two syllables
that differ only in the consonantal onset.
The second group of behavioural methods used
in infant research relies on infants’ looking behaviour. While it may seem counterintuitive to use
infants’ looking behaviour to determine what infants
know about the sound of their mother tongue, these
methods are among the most common for investigating infants’ knowledge about spoken language.
Looking-time experiments rely on the observation
that infants tend to be more attentive to environmental stimuli that they have not seen or heard
before than to familiar ones. As a consequence, they
look longer to novel than to familiar stimuli (9). For
example, in a typical looking-time experiment an
infant sees a visual stimulus on a screen, for example a check-board, and hears simultaneously an
auditory stimulus until the infant’s looking decreases
to a criterion (usually 50%). When the infant has
been habituated, the auditory stimulus changes
while the visual stimulus remains the same. If the
infant discriminates the new sound from the familiar one, it is expected to regain its interest in the
visual display. The difference between the lookingtime to novel and familiar auditory stimuli is taken
as a measure of discrimination.
While simple looking-time measures have successfully been used even in newborns, the methodology requires the experimenter to code the looking
behaviour manually offline. To reduce the manual
work, to avoid subjective biases of the coders and
to obtain a more precise measure for infants’ looking behaviour, today researchers use eye-trackers.
Eye-trackers emit infrared light that is reflected on
the cornea of the eyeball. This reflection of the light
on the cornea is recorded by a sensor that then calculates the exact point of gaze of each eye up to 120
times per second. Because modern eye-trackers no
longer need to be fixed on the subject’s head, the
eye-tracking method allows researchers to adapt
classic looking-time paradigms and behavioural
methods to measure infants’ behavioural reflexes to
environmental stimuli highly precisely in real time.
However, because during the first weeks of life the
human eye goes through considerable changes, it is
only possible to measure eye movements after three
months of age.
Language development in infants
Hearing, Balance and Communication Downloaded from informahealthcare.com by 147.122.60.69 on 07/23/13
For personal use only.
Neuroimaging methods
Because human infants lack fine control and coordination over their motor behaviour, researchers
have increasingly turned to imaging techniques that
measure neural responses to environmental stimuli.
As neuroimaging techniques do not require an overt
behavioural response they are ideally suited for
studying cognitive and linguistic processing in very
young infants. With the advancement of brain imaging techniques, it has become increasingly possible
to investigate the biological foundations of language
with very young infant populations and even with
newborns. In recent years, using advanced imaging
techniques, researchers have therefore also started
to study the specific brain areas that are dedicated
to language perception in newborns and young
infants. This line of research mainly uses two imaging techniques.
Electroencephalography (EEG) is the recording
of electrical activity along the scalp by measuring the
voltage fluctuations resulting from ionic current flows
within the neurons of the brain. EEGs can detect
changes over milliseconds. Electroencephalography
(EEG) and event-related potentials (ERPs) provide
non-invasive methodological techniques that allow
researchers to examine the relation between brain
and behaviour in infancy. Because EEG and ERPs
can be utilized across the entire lifespan, they provide
two of the very few methodological tools that can
directly measure and compare cognitive and linguistic processing at different ages. Even though EEG
and ERPs both reflect the electrical activity of the
brain, and both are collected in a similar manner,
they represent slightly different aspects of brain function: EEG measures the brain’s ongoing electrical
activity, while ERPs show changes in electrical activity in response to a discrete stimulus or event. EEG
provides information regarding the resting state of
the brain, synchrony between regions, or spectral
changes in response to a cognitive event. In contrast,
deflections in the ERP reflect specific aspects of sensory and cognitive processes associated with various
stimuli. Due to the high temporal resolution, ERPs
are well suited for studying the mental processing of
environmental stimuli as they unfold.
However, infant research is increasingly also
interested in understanding brain organization during the first weeks of life. fNIRS (functional nearinfrared spectroscopy) is a non-invasive imaging
method ideally suited to test auditory competence
in newborn infants. fNIRS measures infants’ brain
activity through haemodynamic responses associated with neuron behaviour. It relies on the fact
that when monochromatic light travels through a
medium, some of it is absorbed in the medium,
3
some of it is scattered and some of it is transmitted.
For example, familiar auditory stimuli cause a
decrease of oxyhaemoglobin and novel auditory
stimuli and an increase in the oxyhaemoglobin (11).
The properties of the vascular response measured
using fNIRS are therefore comparable to those
described for the BOLD (blood oxygen level
dependence) effect in fMRI (10). This makes the
methodology highly relevant for studying brain
organization and brain responses in very young
infants who cannot be tested with fMRI. fNIRS
has successfully been used to show the cortical
organization of the newborn brain (10), memory
for words (11) and cortical processing of linguistic
structures (12). For a recent overview of the
methodology see Gervain et al. (13).
Tuning into the language of the environment
Humans start acquiring language from birth (10,12),
if not before (14,15). The auditory system of humans
becomes functional around the 25th week of gestation. This enables the foetus to begin processing
linguistic information before birth. The auditory
information from the environment is filtered and frequencies above 1000 Hz are attenuated (16). This
suggests that they have, at best, limited access to the
acoustic information necessary for discriminating
segmental information, especially consonants. The
foetal hearing experience may thus affect some
aspects of speech perception more than others.
However, the sound that does reach foetuses
provides sufficient information for the perception
of suprasegmental aspects of speech. Importantly,
foetuses also encode speech information into
memory ! especially suprasegmental information.
Both newborns and foetuses demonstrate the ability
to discriminate their native language from a foreign
language (17–20) and their mother’s voice from
another woman’s voice (21,22).
Because most fine-grained segmental information
is filtered out before it reaches the foetus, foetuses can
encode only some suprasegmental characteristics of
speech. It is therefore interesting that human infants
are prepared for language perception the day they are
born. For example, Peña et al. (10) played newborn
human infants speech streams, the same speech
streams played backward, and silence while measuring infants’ brain activity with fNIRS. Backward
speech is the best possible control for speech, since
playing speech backward preserves the physical characteristics of the speech signal, but results in an
acoustic realization that the human vocal tract cannot
produce. Importantly, newborn infants’ brain
responses showed an increased cerebral activity in the
left hemisphere, as with adults, when listening to
Hearing, Balance and Communication Downloaded from informahealthcare.com by 147.122.60.69 on 07/23/13
For personal use only.
4
A. Langus & M. Nespor
speech compared to when listening to backward
speech or silence. These results therefore suggest that
the ability to perceive speech does not require experience with spoken language after birth, and that the
cortical areas responsible for the perception of
spoken language are functional already at birth.
Newborn human infants also show remarkable
abilities in speech perception that they cannot have
developed in utero. During the very first days of life,
newborns distinguish any pair of phonemes attested
in one of the languages of the world (8,23,24). This
ability to discriminate phonemes matures around 28
weeks of gestation as shown also in pre-term infants
(15). Furthermore, it has also been shown that newborns can distinguish the location of word primary
stress: they hear the difference between pápa and
papá or méta and metá (25). The ability to distinguish both phonemes and the phonological realization of lexical stress shows that at birth human
infants are capable of distinguishing all the sounds
used in natural languages. At birth human infants
are thus ready to learn any language to which they
may be exposed.
The question is how infants tune in to the sounds
of their language(s) of exposure. Adult humans are
not particularly good at distinguishing sounds they
are not familiar with. When perceiving sounds from
foreign languages they tend to assimilate these to
sounds that exist in their own language. For example,
a native speaker of French – a language with systematic word final stress – loses the capability of distinguishing stress in other locations. In an experiment
where monolingual native speakers of French were
asked to say if a word they heard was wásuma or
wasúma, they were unable to discriminate the two
stress patterns (26,27). Similarly, monolingual speakers of Japanese ! a language with very restricted
possibilities of consonant clusters ! asked to say if
they heard ebzo or ebuzo were at chance (28).
Infants slowly restrict their discrimination abilities to speech sounds that are systematically present
in their environment. Foreign vowels start being
ignored towards the sixth month of life (29,30) and
foreign consonants towards the ninth month of life
(31). By the first year of life infants have lost the
ability to discriminate phonemes from foreign languages that are not present in their mother tongue.
This focusing on the sounds present in the environment has expressively been called ‘learning by forgetting’ (32): infants learn to ignore sounds that are not
used to distinguish meaning in the language of their
surrounds, or that are not used systematically. Interestingly, they ignore not only the sounds of foreign
languages but also the individual variability of speakers. For example, if certain individuals speak Italian
R
with a uvular r [ ] rather than an alveolar r [r], an
infant successfully learns to ignore this difference.
While it is difficult to sketch the exact time-frame
in which infants narrow down to specific aspects of
spoken language, there is some evidence which suggests that native language influences vocal production already at birth. For example, a recent study
monitored the early sound production of French and
German newborns and found that the melodic contours of the cries of the two groups differed significantly (33). The cries of the French newborns showed
a raising melody while those of the German group
had a falling melody. These differences between newborns in the two linguistic environments suggest that
the influence of the native language begins in utero.
The tuning into the native language may therefore at
least partially be already present in newborns, and
gradually shapes the perception of spoken language
throughout the first year of life.
The basic rhythmic level signals
syllabic structure
Having established that human infants are born with
the ability to tell apart the different segments of the
world’s languages, we can now begin our journey
into the perception of rhythm. The basic rhythmic
level accounts for the different rhythmic classes
attested in the languages of the world. The presence
of rhythmic classes was first proposed by Loyd James
(34) who observed that English sounds very different from Spanish: the first resembles the noise of
messages in the Morse code, the second the sound
of a machine gun (34). There have been several proposals as to which phonetic element is responsible
for these different rhythms. For example, the specific
elements that established isochrony are interstress
intervals in English, and syllables in Spanish (35)
and all languages belong to one of these rhythmic
classes: they are either stress-timed or syllable-timed
(36). A third class was added later to account for the
rhythm of Japanese (37), where the phonological
constituent that establishes isochrony was proposed
to be the mora. However, subsequent work by different phoneticians revealed that isochrony between
different constituents is not present in the speech
signal (38,39).
Which property of the signal is then responsible
for the machine-gun vs. Morse code effect, i.e. for
the undeniable rhythmic difference between English, Dutch or Russian on the one hand, and Spanish, Greek or Italian, on the other? These languages
differ in their syllable structures: stress-timed languages have a greater variety of syllable types than
syllable-timed languages and, as a result, they have
Hearing, Balance and Communication Downloaded from informahealthcare.com by 147.122.60.69 on 07/23/13
For personal use only.
Language development in infants
heavier syllables. Furthermore, in stress-timed languages, unstressed syllables usually have a reduced
vocalic system (sometimes reduced to just one vowel,
schwa), and unstressed vowels are consistently
shorter than stressed vowels, or even absent (40).
These features combine with one another to give the
impression that some syllables are far more salient
than others in stress-timed languages, and that all
syllables tend to be equally salient in syllable-timed
languages (41). It was therefore proposed that the
measures responsible for this basic rhythmic structure are the space occupied by vowels in the speech
stream (V%) and the standard deviation of consonantal intervals (∆C) (3).
This proposal of basic linguistic rhythm where
languages fall into different rhythmic classes on the
basis of the percentage of time occupied by vowels
and on the regularity with which vowels recur – that in
turn predicts the richness of the syllabic repertoire –
has received considerable support from infant studies. For example, French newborns discriminate two
languages such as English and Japanese that belong
to different rhythmic classes, but not languages such
as English and Dutch that belong to the same rhythmic class (20). It can thus be said that newborns
identify the rhythmic class of their language of exposure. Infants fail to discriminate between languages
belonging to the same rhythmic class throughout the
first months of life (42). Only four-month-old bilingual infants exposed to two languages from the same
rhythmic class (Spanish and Catalan) can discriminate the two languages they are actually exposed to
(43). There is thus strong evidence for the proposal
that during the first months of life infants rely primarily on the basic rhythmic level to discriminate
between languages. Telling apart languages from the
same rhythmic class appears to require experience
with the specific languages.
What could infants learn through the identification of the rhythmic class of their native language?
The ability to discriminate languages from different
rhythmic classes shows that infants – in addition to
being able to discriminate between different phonemes – can use quantitative information about consonants and vowels to tell apart languages they may
encounter in their surrounding. These results furthermore suggest that newborn infants may have an idea
of the richness of the syllabic repertoire of their language of exposure. A high %V implies that there are
not so many consonants, and the syllables must be
quite simple. A low %V implies instead many consonants, thus a complex syllabic structure (3). A simple
syllabic structure is in turn predictive of word length
since the number of possible monosyllables that could
be used as alone-standing words is restricted. This is
the case, for example, of Japanese, where most words
5
are polysyllabic. Languages with complex syllabic
structure – such as English or Dutch – are instead
rich in monosyllables. The identification of the rhythmic class could thus also offer a bias as to the mean
length of common words and thus provide a valuable
aid to speech segmentation (4).
The sensitivity to this basic level of linguistic
rhythm in early stages of cognitive development is
also important because there is a partial division of
labour between vowels and consonants. While the
main role of consonants concerns the lexicon, the
main role of vowels is that of allowing the identification of the rhythmic class as well as of the development of a bias as to the mean length of common
words (44). For example, in an artificial speech experiment adult participants prefer to use consonants to
vowels in word identification (45) and they tend to
use relationships among vowels to extract generalizations while disregarding the relationships among consonants for the same purpose (46).These asymmetries
are not by-products of low-level acoustic differences
between consonants and vowels (47,48), but result
from the categorical representations of consonants
and vowels themselves. This bias for using consonants
for word learning and vowels for rule learning is
found in infants by at the latest 12 months of age
(49). The ability to perceive linguistic rhythm – to
discriminate, represent and use phonemes categorically – from the earliest stages of language acquisition
may therefore be necessary for the acquisition of
words and rules of the mother tongue.
Phrasal rhythm signals word order
At a hierarchical level higher than that of the alternation between vowels and consonants, rhythm is
determined by the alternation of stresses. It is important to note that the different levels of the prosodic
hierarchy are organized so that lower levels are
exhaustively contained into higher ones (50). This is
best exemplified by considering the prosodic constituents most relevant for the present paper: the
Phonological Phrases. The Phonological Phrase
extends from the left edge of a phrase to the right
edge of its head in head-complement languages; and
from the left edge of a head to the left edge of its
phrase in complement-head languages (1). While the
number of Phonological Phrases contained in an
Intonational Phrase that immediately governs them
may vary, Phonological Phrases never straddle Intonational Phrase boundaries; Phonological Phrases
are exhaustively contained in Intonational Phrases.
The most well known use for Phonological Phrases
is for speech segmentation. Because syntactic structure is automatically mapped onto prosodic structure
during speech production (1), many prosodic cues
Hearing, Balance and Communication Downloaded from informahealthcare.com by 147.122.60.69 on 07/23/13
For personal use only.
6
A. Langus & M. Nespor
signal syntactic and word boundaries. For example,
among the most common cues for Phonological
Phrase boundaries is final lengthening (51–54). Adult
listeners use phonological phrase boundaries to
identify where words begin or end (55). Similarly,
10-month-old infants can use phonological phrase
boundaries to find words from continuous speech
(56). Furthermore, because prosodic boundaries are
necessary for infants to detect rule-like regularities
from continuous speech (57), phonological phrase
boundaries may also help infants to extract rules from
different levels of the prosodic hierarchy (58).
However, the systematic recurrence of Phonological Phrases is also important because the location
as well as the physical manifestation of stress in Phonological Phrases give a cue to the word order of a
language. Of the six logically possible orders of Subject (S), Object (O) and Verb (V), the basic word
order of the great majority of the languages of the
world (76%) is either SOV – as in Turkish, Japanese
or Basque ! or SVO – like English, Hausa or Greek
(59). Nespor et al. (6) proposed that the stress in
Phonological Phrases differs systematically between
SVO and SOV languages. Languages with the SVO
order have stress in final position (e.g. on the Noun
in a Verb-Noun sequence) manifested mainly through
duration, i.e. prominent syllables are longer than less
prominent ones. Languages with the SOV order have
stress in initial position (e.g. on the Noun in a NounVerb sequence) manifested mainly through pitch and
intensity, i.e. prominent syllables are more intense
and have a higher pitch than less prominent ones.
Given the uniformity of word order across constituents of different types within one language, could
the specific word order of the language of exposure be
inferred from the manifestation of phrasal rhythm in
the prelexical period? To begin answering this question,
French 6- and 12-week-old infants were exposed to
French (SVO) and Turkish (SOV) sentences that differed solely on the location of relative prominence
within phonological phrases (5). Turkish and French
were chosen in this study because their syllabic structure is similar in complexity and because they both
have word final stress. To eliminate the possible effect
of segments
– e.g. French but not Turkish has a uvular
R
‘r’ [ ]; Turkish but not French has a high-unrounded
posterior vowel – all sentences were resynthesized so
as to have the same segments for each category. The
findings of the study show that infants at this young
age could discriminate Turkish and French sentences
exclusively on the basis of phrasal rhythm, highlighting
that phonological phrase level rhythm is perceived
already during the first months of life.
If infants can discriminate two languages solely on
the basis of their phrasal prominence, are they also able
to hear if in a sequence of alternating weak and strong
beats, the strong beat is initial or final in its group? This
puzzle could be solved if a perceptual mechanism
could indicate if a strong element is initial or final. A
perceptual mechanism first proposed to account for
grouping in music (60–62) – the Iambic-Trochaic Law
(ITL) – was proposed to be responsible for the different acoustic manifestation of phrasal stress depending
on whether it is initial or final in its phrase (6). The
ITL states that a sequence of sounds that alternates in
duration is grouped iambically (weak-strong), while a
sequence of sounds that alternates in intensity – and
pitch for speech at the phonological phrase level –
is grouped trochaically (strong-weak). Whether a
language has the Verb-Object or the Object-Verb order
is thus signalled by phrasal rhythm.
Adults and seven-month-olds were tested for
their memory of sequences of syllables that alternate
in either pitch or duration (63). Exposed to flat syllables in the test phase, adults were better at remembering pairs of syllables that during familiarization
had short syllables preceding long syllables, or highpitched syllables preceding low-pitched syllables.
Instead, infants familiarized with syllables alternating
in pitch, showed a preference to listen to pairs of
syllables that had high pitch in the first syllable.
However, no preference was found when the familiarization stream alternated in duration. One possible
explanation for this asymmetry is related to the syntactic difference between SVO and SOV languages.
It is widely accepted that SVO is syntactically
unmarked and that all languages are at an underlying
syntactic level SVO. The surface SOV order would
then be result of movement (64,65). External confirmation for this theory comes from historical change:
change in word order is unidirectional from SOV to
SVO (66, 67). It is therefore possible that infants are
sensitive early on only to the prosodic profile that
indicates that the order of their language of exposure
is not the unmarked one.
Taken together, there is strong evidence from language acquisition that the perception of linguistic
rhythm is one of the first aspects of language young
infants perceive and process. Because rhythm is, at
different levels of the rhythmic hierarchy, carried by
the core elements of the speech signal, it is helpful
for a variety of tasks that the young language learners
face: from speech segmentation to learning one of
the basic properties of the syntactic structure of the
language to be acquired.
Sensitivity to speech sounds by cochlear
implanted infants
While studies on language acquisition zoom in on
the factors that influence speech perception and
processing, it is often difficult to disentangle which
Hearing, Balance and Communication Downloaded from informahealthcare.com by 147.122.60.69 on 07/23/13
For personal use only.
Language development in infants
abilities require early exposure to linguistic input. As
newborns are able to discriminate all the sounds
found in the world languages, listening to speech
in early infancy would appear to primarily help to
narrow their cognitive abilities and tune into their
mother tongue. However, because native language
acquisition is constrained by critical periods – or the
window of opportunity – during which native competence of a language can be acquired (68), it is also
likely that this narrowing effect of speech input is
sensitive to the specific stage of language acquisition.
One source of valuable information about how listening to speech modulates our cognitive abilities is
individuals who are born deaf and gain hearing
through cochlear implants.
Speech intelligibility in cochlear implanted individuals is directly related to the quality of the prosody
they hear (69). There is some evidence which suggests
that cochlear implanted children perform significantly
worse than normal controls in the perception of
amplitude, pitch and temporal structure of spoken
language (70). Studies in languages such as Chinese
and Thai, where the lexical meaning of words is modulated by tones, show that the main problem for
cochlear implanted children lies in the perception of
pitch (71). This is also evident from studies which
show that cochlear implanted children have also problems in interpreting intonation and its relation with
the elocutionary force of a sentence, e.g. distinguishing a statement from a question. As a consequence
they also have a significantly poorer performance than
normally-hearing individuals in the perception and
production of emotional prosody (72).
It remains an open question whether prosodic
processing can only emerge if the speech input is
present during the window of opportunity – the critical period during which language can be acquired,
or at least can be acquired natively. However, because
rhythm interprets the grammatical structure of sentences at different levels of the prosodic hierarchy,
the inability to process prosody may cause serious
difficulties in acquiring language in individuals who
receive a cochlear implant after a certain age. On the
one hand, the research described above may be useful to diagnose hearing impairments in infants and
young children. On the other hand, it may help to
find ways to train the perception and production of
prosody in cochlear implanted individuals – both
children and adults. One possibility, albeit an unexplored one, may be to complement auditory training
with training in the perception and production of
rhythm in a non-acoustic modality, e.g. in the visual
modality. For example, it has been shown that
the Iambic-Trochaic Law determines grouping of
sequences also in vision, where adult participants
can group sequences of shapes according to visual
7
frequency, intensity and duration as in the auditory
domain (73).
Conclusions
From the very beginning of life, humans are sensitive
to all consonantal and vocalic distinctions present in
the different languages of the world. Their sensitivity
to prosody – rhythm and intonation – appears to
emerge already in the last weeks of gestation. Knowledge of vowels and consonants, that carry the
basic level of linguistic rhythm, is necessary to learn
the lexicon. There is evidence that consonants are
privileged in learning words because of their relative
stability. In contrast, vowels, because of their variability, provide more information about the grammatical structure of languages. Specifically, the
percentage of time vowels occupy in the speech
stream can give a cue to the size of the syllabic repertoire of the language of exposure and thus to the
average length of common words. Because vowels are
the main carrier of prosody they also give a cue to
one of the basic syntactic properties of language !
word order. Cochlear implanted children have been
shown to be especially impaired in prosody in general
and in the perception – and thus the production – of
tonal distinctions, in particular. This may suggest
that the prosodic processing of speech, especially
of intonation, is a precarious ability that requires
experience with speech from the earliest stages of
development.
Declaration of interest: The authors report no
conflicts of interest. The authors alone are responsible for the content and writing of the paper.
The research leading to these results has
received funding from the European Research
Council under the European Union’s Seventh
Framework Programme (FP7/2007-2013) / ERC
grant agreement n° 269502 (PASCAL).
References
1. Nespor M, Vogel I. Prosodic Phonology. 2nd edition 2007.
Dordrecht: Foris; 1986.
2. Nespor M. On the rhythm parameter in phonology. In: Roca
I, editor. Logical issues in language acquisition. Dordrecht:
Foris; 1990. p. 1–19.
3. Ramus F, Nespor M, Mehler J. Correlates of linguistic rhythm
in the speech signal. Cognition. 1999;73: 265–92.
4. Nespor M, Shukla M, Mehler J. Stress-timed vs. syllable
timed languages. In: van Oostendorp M, Ewen CJ, Hume E,
Rice K, editors. The Blackwell Companion to Phonology.
Chichester: Wiley-Blackwell; 2011. p. 1147–59.
5. Nespor M, Guasti MT, Christophe A. Selecting word order:
the Rhythmic Activation Principle. In: Kleinhenz U, editor.
8
6.
7.
8.
9.
Hearing, Balance and Communication Downloaded from informahealthcare.com by 147.122.60.69 on 07/23/13
For personal use only.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
A. Langus & M. Nespor
Interfaces in Phonology. Berlin: Akademie Verlag; 1996.
p. 1–26.
Nespor M, Shukla M, van de Vijver R, Avesani C,
Schraudolf H, Donati C. Different phrasal prominence
realization in VO and OV languages. Lingue e Linguaggio.
2008;7:1–28.
Siqueland ER, DeLucia CA. Visual reinforcement of
non-nutritive sucking in human infants. Science. 1969;165:
1144–6.
Eimas PD, Siqueland ER, Jusczyk P, Vigorito J. Speech
perception in infants. Science. 1971;171:303–6.
Gottlieb G. Krasnegor NA. Measurement of Audition and
Vision in the First Year of Postnatal Life: A Methodological
Overview. Connecticut: Ablex; 1985.
Peña M, Maki A, Kovačić D, Dehaene-Lambertz G,
Koizumi H, Bouquet F, Mehler J. Sounds and Silence: an
optical topography study of language recognition at birth.
Proc Nat Acad Sci USA. 2003;100:11702–5.
Benavides-Varela S, Hochmann JR, Macagno F, Nespor M,
Mehler J. Newborns’ brain activity signals the origin of word
memories. Proc Nat Acad Sci USA. 2012;109:17908–13.
Gervain J, Macagno F, Cogoi S, Peña M, Mehler J. The
neonate brain detects speech structure. Proc Nat Acad Sci
USA. 2008;105:14222–7.
Gervain J, Mehler J, Werker JF, Nelson CA, Csibra G,
Lloyd-Fox S, et al. Near-infrared spectroscopy: a report from
the McDonnell infant methodology consortium. Dev Cogn
Neurosci. 2011;1:22–46.
Peña M, Pittaluga E, Mehler J. Language acquisition in
premature and full-term infants. Proc Nat Acad Sci USA.
2010;107:3823–8.
Mahmoudzadeh M, Dehaene-Lambertz G, Fournier M,
Kongolo G, Goudjil S, Dubois J, et al. Syllabic discrimination
in premature human infants prior to complete formation of
cortical layers. Proc Nat Acad Sci USA. 2013;110:4431–2.
Lecanuet JP, Granier-Deferre C, Jacquet AY, Capponi I,
Ledru L. Prenatal discrimination of a male and female voice
uttering the same sentence. Early Dev Parent. 1993;2:
217–28.
Kisilevsky B, Hains S, Brown C, Lee C, Cowperthwaite B,
Stutzman S, et al. Foetal sensitivity to properties of maternal
speech and language. Infant Behav Dev. 2009;32:59–71.
Mehler J, Jusczyk P, Lambertz G, Halsted N, Bertoncini J,
Amiel-Tison C. A Precursor of Language Acquisition in
Young Infants. Cognition.1988;29:143–78.
Moon C, Panneton-Cooper R, Fifer WP. Two-day-olds prefer
their native language. Infant Behav Dev. 1993;16:495–500.
Nazzi T, Bertoncini J, Mehler J. Language Discrimination by
Newborns: Toward an Understanding of the Role of Rhythm.
J Exp Psychol: Hum Percept Perform. 1998;24:1–11.
DeCasper AJ, Fifer WP. Of human bonding: Newborns prefer
their mothers’ voices. Science. 1980;208:1174–6.
Kisilevsky B, Hains S, Lee K, Xie X, Huang H, Zhang K,
Wang Z. Effects of experience on foetal voice recognition.
Psychobiol Sci. 2003;14:220–4.
Eimas PD, Miller JL. Contextual effects in infant speech
perception. Science. 1980;209:1140–1.
Dehaene-Lambertz G, Peña M. Electrophysiological evidence
for automatic phonetic processing in neonates. NeuroReport.
2001;12:3155–8.
Sansavini A, Bertoncini J, Giovanelli G. Newborns discriminate the rhythm of multisyllabic stressed word. Dev Psychol.
1997;33:3–11.
Cutler A, Mehler J, Norris D, Segui J. The Syllable’s Differing
Role in the Segmentation of French and English. J Mem
Lang. 1986;25:385–400.
27. Cutler A, Mehler J, Norris D, Segui J. The Monolingual
Nature of Speech Segmentation by Bilinguals. Cogn Psychol.
1992;24:381–410.
28. Dupoux E, Kakehi K, Hirose Y, Pallier C, Mehler J.
Epenthetic vowels in Japanese: a perceptual illusion? J Exp
Psychol: Hum Percept Perf .1999;25:1568–78.
29. Kuhl PK. Human adults and human infants show a ‘perceptual magnet effect’ for the prototypes of speech categories,
monkeys do not. Percept Psychophys.1991;50:93–107.
30. Kuhl PK, Williams KA, Lacerda F, Stevens KN, Lindblom B.
Linguistic experience alters phonetic perception in infants by
six months of age. Science. 1992;255:606–8.
31. Werker JF, Tees RC. Cross-language speech perception:
evidence for perceptual reorganization during the first year
of life. Infant Behav Dev. 1984;7:49–63.
32. Mehler J, Dupoux E. Naître Humain. Paris: Odile Jacob;
1990.
33. Mampe B, Friederici A, Christophe A, Wermke K. Newborns’
cry melody is shaped by their native language. Curr Biol.
2009;19:1994–7.
34. Lloyd James A. Speech Signals in Telephony. London; 1940.
35. Pike KL. The Intonation of American English. Ann Arbor:
University of Michigan Press; 1945.
36. Abercrombie D. Elements of General Phonetics. Chicago:
Aldine; 1967.
37. Ladefoged P. A Course in Phonetics. New York: Harcourt
Brace Jovanovich; 1975.
38. Borzone de Manrique AM, Signorini A. Segmental durations
and the rhythm in Spanish. J Phon. 1983;11:117–28.
39. Wenk B, Wiolland F. Is French really syllable-timed? J Phon.
1982;10:193–216.
40. Bertinetto PM. Strutture prosodiche dell’ italiano. Accento,
quantità, sillaba, giuntura, fondamenti metrici. Firenze:
Accademia della Crusca; 1981.
41. Dauer R. Phonetic and phonological components of language
rhythm. Paper presented at the XIth International Congress
of Phonetic Sciences, Tallinn; 1987.
42. Christophe A, Morton J. Is Dutch native English? Linguistic
analysis by two-month-olds. Dev Sci. 1998;1:215–9.
43. Bosch L, Sebastian-Galles N. Evidence of Early Language
Discrimination Abilities in Infants From Bilingual Environments. Infancy. 2001;2:29–49.
44. Nespor M, Peña M, Mehler J. On the Different Roles of
Vowels and Consonants in Speech Processing and Language
Acquisition. Lingue & Linguaggio 2003;2:203–31.
45. Bonatti L, Peña M, Nespor M, Mehler J. Linguistic Constraints on Statistical Computations: The Role of Consonants
and Vowels in Continuous Speech Processing. Psychol Sci.
2005;16:451–9.
46. Toro JM, Bonatti LL, Nespor M, Mehler J. Finding words
and rules in a speech stream: functional differences between
vowels and consonants. Psychol Sci. 2008;19:137–44.
47. Toro JM, Shukla M, Nespor M, Endress A. The quest
for generalizations over consonants: asymmetries between
consonants and vowels are not the by-product of acoustic
differences. Percept Psychophys. 2008;70:1515–25.
48. New B, Nazzi T. The time-course of consonant and vowel
processing during word recognition. Lang Cogn Process.
2012;1:1–15.
49. Hochmann JR, Benavides S, Nespor M, Mehler J.
Consonants and vowels: different roles in early language
acquisition. Dev Sci. 2011;14:1445–58.
50. Selkirk E. Phonology and syntax: the relation between sound
and structure. Cambridge: The MIT Press; 1984.
51. Beckman M, Pierrehumbert J. Intonational structure in
Japanese and English. Phonology Yearbook. 1986;3:15–70.
Hearing, Balance and Communication Downloaded from informahealthcare.com by 147.122.60.69 on 07/23/13
For personal use only.
Language development in infants
52. Cooper WE, Paccia-Cooper J. Syntax and Speech.
Cambridge: Harvard University Press; 1980.
53. Klatt DH. Linguistic uses of segmental duration in English:
acoustic and perceptual evidence. J Acoust Soc Am. 1976;
59:1208–21.
54. Wightman CW, Shattuck-Hufnagel S, Ostendorf M,
Price PJ. Segmental durations in the vicinity of prosodic
phrase boundaries. J Acoust Soc Am. 1992;91:1707–17.
55. Christophe A, Peperkamp S, Pallier C, Block E, Mehler J.
Phonological phrase boundaries constrain lexical access: I.
Adult data. J Mem Lang. 2004;51:523–47.
56. Gout A, Christophe A, Morgan J. Phonological phrase
boundaries constrain lexical access: II. Infant data. J Mem
Lang. 2004;51:547–67.
57. Peña M, Bonatti L, Nespor M, Mehler J. Signal-Driven
Computations in Speech Processing. Science. 2002;298:
604–7.
58. Langus A, Marchetto E, Bion RAH, Nespor M. Can prosody
be used to discover hierarchical structure in continuous
speech? J Mem Lang. 2012;66:285–306.
59. Dryer MS. The order of subject, object and verb.
In: Haspelmath M, Dryer MS, Gil D, Comrie B, editors.
The World Atlas of Language Structures. Oxford: Oxford
University Press. 2005. p. 330–3.
60. Bolton T. Rhythm. Am J Psychol. 1894;6:145–238.
61. Cooper G, Meyer L. The Rhythmic Structure of Music.
Chicago: University of Chicago Press; 1960.
62. Woodrow H. Time perception. In: Stevens S, editor.
Handbook of Experimental Psychology. New York: Wiley;
1951. p. 1224–36.
9
63. Bion RAH, Benavides-Varela S, Nespor M. Acoustic markers
of prominence influence infants’ and adults’ segmentation of
speech sequences. Lang Speech. 2011;54:123–40.
64. Kayne RS. The Antisymmetry of Syntax. Cambridge: MIT
Press; 1994.
65. Chomsky N. Minimalist Program. Cambridge: MIT Press;
1995.
66. Newmeyer FJ. On the reconstruction of ‘Proto-world’ word
order. In: Knight C, Hurford JR, Studdert-Kennedy M,
editors. The Evolutionary Emergence of Language: social
function and the origins of linguistic form. Cambridge:
Cambridge University Press; 2000.
67. Gell-Mann M, Ruhlen M. The origin and evolution of word
order. Proc Nat Acad Sci U S A. 2011;108:17290–5.
68. Lenneberg EH. Biological Foundations of Language. Wiley;
1967.
69. Chin SB, Bergeson TR, Phan J. Speech intelligibility and
prosody production in children with cochlear implants.
J Commun Disord. 2012;45:355–66.
70. Meister H, Tepeli D, Wagner P, Hess W, Walger M, von Wedel
H, Lang-Roth R. Experiments on prosody perception with
cochlear implants. HNO. 2007;55:264–70.
71. Ciocca V, Francis AL, Aisha R, Wong L. The perception of
Cantonese lexical tones by early-deafened cochlear implantees, J Acoust Soc Am. 2002;111:2250–6.
72. Nakata T, Trehub SE, Kanda Y. Effect of cochlear implants
on children’s perception and production of speech prosody.
J Acoust Soc Am. 2012;131:1307–14.
73. Peña M, Bion RAH, Nespor M. How modality specific is the
Iambic-Trochaic Law? Evidence from vision. J Exp Psychol:
Learn Mem Cogn. 2011;37:1199–1208.